Intermetallics 14 (2006) 1226e1230

www.elsevier.com/locate/intermet

Phase equilibria and phase transformations in the Ti-rich

corner of the FeeNieTi system
P. Riani a, G. Cacciamani a,*, Y. Thebaut b, J. Lacaze b
a
Dipartimento di Chimica e Chimica Industriale, Università di Genova, via Dodecaneso 31, I-16146 Genova, Italy
b
CIRIMAT, UMR CNRS 5085, ENSIACET, F-31077 Toulouse Cedex 4, France
Received 5 September 2005; received in revised form 11 September 2005; accepted 15 September 2005
Available online 30 March 2006

Abstract

While the main features of the FeeNieTi system are well known at low Ti content, literature review of the Ti-rich corner revealed incon-
sistencies between experimental reports. This investigation presents new experimental results, defined to remove the uncertainties concerning
melting behavior and solid-state phase equilibria of the (Ni,Fe)Ti2 phase with the adjacent (Fe,Ni)Ti (B2, CsCl-type structure) and b-Ti (A2,
W-type) phases. Six samples have been prepared and examined by differential thermal analysis performed in yttria and alumina crucibles,
and by scanning electron microscopy in the as-cast state as well as equilibrated at 900
C.
Ó 2006 Elsevier Ltd. All rights reserved.

Keywords: A. Ternary alloy systems; A. Intermetallics, miscellaneous; B. Phase diagrams; D. Microstructure

1. Introduction
FeeNieTi is an important ternary system relevant to differ-
ent kinds of materials: from special steels and Ni-based super-
alloys to alloyed aluminides, from Ti-based corrosion-resistant
alloys to shape memory alloys and amorphous materials. Fee
NieTi is also a sub-system of the AleFeeNieTi quaternary
system, which is the core system of COST 535 ‘‘THALU’’,
a European project aimed at the investigation of the constitu-
tional properties of alloyed aluminides by means of experi-
ments, thermodynamic and ab initio calculations. A review
of the ternary FeeNieTi system provided by Gupta [1] was
mainly based on the works by Vogel and Wallbaum [2], Speich
[3] and Van Loo et al. [4]. From the first two of these works
and the limiting binary systems, Gupta proposed a schematic
liquidus projection. It is shown in Fig. 1 according to the revi-
sion by Cacciamani et al. [5] in the Ti-rich corner, as discussed
and summarized below. The binary invariant points selected to
* Corresponding author. Tel.: þ39 010 353 6152; fax: þ39 010 362 5051.
E-mail address: cacciamani@chimica.unige.it (G. Cacciamani).
draw this figure agree with the assessments of Lee [6] for the
FeeNi system, Kumar [7] for the FeeTi system and Bellen
et al. [8] for the NieTi one. Van Loo et al. [4] provided the
900
C isothermal section shown in Fig. 2. This figure indi-
cates that every binary compound, NiTi2, (Fe,Ni)Ti, Fe2Ti
and Ni3Ti, presents significant solubility for the third element
and that no stable ternary compound occurs at that tempera-
ture. As a matter of fact, no ternary compound was found in
this system at any temperature. It is worth mentioning the
complete solubility of Fe and Ni in the ordered B2-(Fe,Ni)Ti
phase which was previously evidenced by Dudkina and
Kornilov [9]. Abramycheva et al. [10] realized a study similar
to the one carried out by Van Loo et al. [4] but at 1000
C
from which they drew an isothermal section that is mostly sim-
ilar to Fig. 2 apart from the presence of a liquid field around
point e5 in Fig. 1 (which is the binary eutectic liquid/b-Ti/
NiTi equilibrium at 942
C).
In a re-assessment of his previous review, Gupta [11] ac-
counted for the work of Alisova et al. [12] who provided three
vertical sections of the diagram in the Ti-rich region. These
sections were selected with Fe:Ni equal to 3:1, 1:1 and 1:3,
respectively, and suggested that the peritectic reaction giving

0966-9795/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.intermet.2005.09.011
 P. Riani et al. / Intermetallics 14 (2006) 1226e1230 1227

of Ti and involved structural analysis of homogenized samples

and as-cast materials as well as differential thermal analysis of
the first ones.

2. Experimental details

Several alloys with composition along the FeTi2eNiTi2 line

were prepared by arc melting small pieces of the pure
elements (99.99 wt% Fe, 99.99 wt% Ni, and 99.98 wt% Ti)
in an argon atmosphere. The alloys were then sealed in
quartz ampoules filled with argon, heat treated for 15 days at
900
C to reach equilibrium and then quenched in water.
The samples were observed by optical (Leica DM 4000M)
and scanning electron microscopies (SEM, Cambridge Stereo-
scan 200) and analysed by energy dispersive spectroscopy
(EDS, Oxford INCA Energy 300) to evaluate the composition
of the coexisting phases and of the overall samples. The mi-
Fig. 1. Liquidus projection of the FeeNieTi system according to Cacciamani
crostructure of the alloys was investigated after surface polish-
et al. [5].
ing by using SiC papers and diamond paste down to 1/4 mm
particle size. SEM imaging in backscattered electron mode
NiTi2 occurs only in the section at Fe:Ni ¼ 1:3 closest to the as well as EDS analyses were performed on unetched samples.
NieTi side. Accordingly, Gupta [11] located the four-phase Quantitative EDS analyses were carried out using an acceler-
equilibrium between liquid, b-Ti, NiTi2 and B2-(Fe,Ni)Ti at ation voltage of 20 kV and a counting time of 50 s using a
a very low Fe content. The vertical sections determined by Ali- cobalt standard in order to monitor beam current, gain and
sova et al. [12] are, however, in contradiction with the isother- resolution of the spectrometer. Apparent compositions were
mal sections by Van Loo et al. [4] and by Abramycheva et al. finally corrected for ZAF (atomic number, absorption and fluo-
[10], as effectively noticed by these latter authors and dis- rescence) effects using pure elements as standards. The stan-
cussed by Cacciamani et al. [5]. Finally, Efimenko et al. dard deviation for each element was estimated at 0.5 at.%
[13] observed experimentally that NiTi2 precipitates from for phase analysis and at 1.0 at.% for global analysis.
melts with much higher Fe contents than suggested by Abra- Small part of the samples was then cut out for differential
mycheva et al. [10]. In order to remove this discrepancy, the thermal analysis (DTA, SETARAM SETSYS) and the remain-
present work was undertaken on a series of alloys at 66 at.% ing part was re-melted and solidified by natural cooling in the
arc furnace for observation of the as-cast microstructure. The
composition of the phases in the as-cast state was also evalu-
Ti ated by means of EDS (PGT analyzer) on an LEO scanning
electron microscope, under experimental conditions very sim-
β−Ti ilar to those listed above except that the counting time was 100 s.
20 80
According to the PGT software, the standard deviation in this
case was about 0.5 at% for Ni and 0.8 at% for Fe and Ti.
Fe (at. ) NiTi2 DTA experiments were first performed in yttria crucibles
40 (provided by MaTeck GmbH, Jülich, Germany) with the
60
Ti (at. ) hope to avoid extensive reaction between the alloy and the cru-
(Fe,Ni)Ti cible. Before each experiment, the DTA cell was twice evacu-
60
40 ated and filled with high purity argon. Runs were performed at
Fe2Ti
a scanning rate of 15 K/min under a continuous flux of argon.
Ni3Ti No exothermic reaction was observed after melting, as it is the
80
20 case when alumina crucibles are used. However, it was noted
Fe that the sample stuck to the wall of the crucible, thus showing
(Fe,Ni) some limited reaction with it. Moreover, WDS analysis (CA-
Fe Ni
20 40 60 80
MECA SX-50 microprobe) of a sample which could be sep-
Ni (at. ) arated from the crucible showed extensive pollution with
oxygen. It was thus decided to consider only the data for start
Fig. 2. FeeNieTi isothermal section at 900
C after Van Loo et al. [4]. Solid of melting of the alloys and it was checked that similar data
circles represent phase compositions according to the accepted assessments of could be obtained with standard alumina crucibles. All the
the binary systems. Empty circles are experimental data read from the
provided figure [4]: they are linked with tie-lines drawn with dots. Three-phase DTA records were thus carried out twice. Before and after
triangles shown with interrupted lines as well as single-phase fields were the whole series of recording, the DTA cell was calibrated us-
guessed by Van Loo et al. [4]. ing melting temperature Tm of pure Ag as illustrated with the
1228 P. Riani et al. / Intermetallics 14 (2006) 1226e1230

the basis of Fig. 2, it was easy to establish that phase 1 corre-

Fig. 3. DTA records obtained on heating pure Ag at various scanning rates.
The tabulated melting temperature Tm of Ag is indicated.
runs at 5 and 15 K/min shown in Fig. 3. As expected, an in-
creased scanning rate shifts more and more the melting signal
with respect to the tabulated temperature.
3. Results
3.1. Homogenized alloys
The list of the samples prepared and the results of the phase
analysis are reported in Table 1. Except for alloy A which
contained three phases, all other homogenized alloys showed
a two-phase microstructure. The minor phase, denoted as
phase 1 in Table 1, has a lower Ti content than phase 2 and
presents remaining outlines of dendritic solidification in all
samples, as illustrated in Fig. 4 for sample C. In alloy A, a third
phase shows up with even higher Ti content than phase 2. On
sponds to B2-(Fe,Ni)Ti, phase 2 to (Fe,Ni)Ti2 and phase 3 to
the A2 b-Ti solid solution. The dendritic shape adopted by
phase 1 shows that it is the primary phase formed upon solid-
ification for all six alloys investigated. Except for alloy A,
(Fe,Ni)Ti2 forms incongruently from liquid and B2-(Fe,Ni)Ti,
as it is also the case in the binary NieTi system. In the case of
alloy A, the three-phase microstructure observed after homog-
enization could not be interpreted so straightforwardly.
In Fig. 5 the overall compositions of the alloys (solid cir-
cles) and those of the individual phases (solid squares) are
plotted. Solid diamonds on the sides represent the composition
of the binary phases at 900
C according to the accepted bi-
nary phase diagrams (see Ref. [5]). Tie-lines for the two-phase
samples and tie-triangle for sample A are also drawn. These
equilibria are very similar to those determined by Van Loo
et al. [4] and reproduced in Fig. 2, consisting in a three-
phase equilibrium b-Ti/(Fe,Ni)Ti2/B2 close to the FeeTi
side and an extended (Fe,Ni)Ti2/B2 two-phase field towards
the NieTi side.
DTA peaks recorded on heating of homogenized alloys
AeF are illustrated in Fig. 6. Owing to the smooth shape
of the peaks, the solidus temperature was estimated at the in-
tersection of the extrapolated base line with the back-extrap-
olated peak rise shown with interrupted lines in the figure.
According to the Ag calibration in Fig. 3, the evaluated tem-
peratures were decreased by 3
C. Fig. 7 shows the experi-
mental solidus temperatures as a function of the iron content
of the alloy. The reproducibility of the records was excellent,
with the highest difference in the evaluation of the solidus
temperature amounting to a few degrees in the case of alloy
D (15 at.% of Fe). For this alloy, the two dots representing
the solidus temperature can be distinguished in Fig. 7, in
all other cases they cannot. In the same figure, assessed tem-
peratures of the binary reactions have been added and they
show that the present set of results is highly consistent with
them. The evolution of the solidus temperature suggests a max-
imum at about 1030
C and 15e20 at% Fe and a minimum

Table 1
Compositions and phase analyses of the homogenized (Homog.) and as-cast alloys determined by EDS
Sample Sample composition Sample state First phase: B2-(Fe,Ni)Ti Second phase: (Fe,Ni)Ti2 Third phase: A2 b-Ti Fourth phase:
Fe/Ni/Ti (at.%) Fe/Ni/Ti (at.%) Fe/Ni/Ti (at.%) Fe/Ni/Ti (at.%) Fe/Ni/Ti (at.%)
A 29.0/4.0/67.0 Homog. 44.0/4.0/52.0 26.0/6.5/67.5 18.0/1.5/80.5
As-cast 43.5/3.8/52.7 22.5/6.5/71.0 20.2/1.9/77.9 33.3/4.3/62.4
B 25.0/8.0/67.0 Homog. 41.5/6.5/52.0 23.5/9.5/67.0
As-cast 37.5/10.2/52.3 22.6/9.9/67.5 17.4/3.6/79.0
C 20.0/13.5/66.5 Homog. 35.0/13.0/52.0 19.0/13.5/67.5
As-cast 30.4/16.6/53.0 18.8/13.8/67.4 12.7/5.3/82.0
D 14.5/19.0/66.5 Homog. 25.0/23.0/52.0 14.0/18.5/67.5
As-cast 23.4/24.2/52.4 14.8/17.9/67.3 7.9/8.9/83.2
E 10.0/24.0/66.0 Homog. 16.5/31.5/52.0 9.5/23.5/67.0
As-cast 16.0/31.4/52.6 8.2/24.1/67.7 4.6/17.6/77.8
F 5.0/28.5/66.5 Homog. 7.5/41.0/51.5 4.5/28.0/67.5
As-cast 6.7/41.2/52.1 5.0/27.8/67.2
 P. Riani et al. / Intermetallics 14 (2006) 1226e1230 1229

Fig. 4. Backscattered electron image of the microstructure of alloy C after

homogenization at 900
C. The brighter phase, lower in Ti, shows typical
dendritic outline and is associated with B2-(Fe,Ni)Ti while the darker one is
(Fe,Ni)Ti2.
at about 1025
C and 25 at% Fe. The minimum could be as-
sociated with the invariant equilibrium between liquid, b-Ti,
NiTi2 and B2-(Fe,Ni)Ti which should thus be quite close to
the TieFe side.
3.2. As-cast samples
The microstructures of the as-cast samples show different
aspects when compared to the equilibrated alloys. Alloy A
apparently presents four phases (Fig. 8(a)), alloys BeE show
three phases (Fig. 8(b)), and alloy F shows only two phases.
In Table 1, the phases have been sorted according to their Ti
content, except for the fourth phase in sample A. It can be
seen that the data for phases 1e3 in Fig. 5 (empty squares),
Fig. 6. DTA records for alloys AeF obtained on heating at 15 K/min. The
curves were shifted arbitrarily along the ordinate axis for better clarity. The
solidus temperatures have been estimated by extrapolation as shown in the fig-
ure with interrupted lines. Before plotting on Fig. 7, the values here indicated
diminished by 3
C to account for heating rate effect (see text).
are very close to the data obtained on homogenized samples
for phase 1 corresponding to B2-(Fe,Ni)Ti and phase 2 cor-
responding to (Fe,Ni)Ti2. The measurements on phase 3 (b-
Ti) are well located between the two extreme values on the
binary sides apart from the one obtained from alloy E which
appears at too high Fe level. As expected, it is seen that, for
alloy A, the Ni content of b-Ti is slightly lower in the
sample homogenized at 900
C than in the as-cast alloy be-
cause the solubility of Ti in this phase decreases with
temperature.
As for the homogenized samples, the B2 phase appears in
every case as a dendritic primary phase. Moreover, in alloys

1100
eutectic liquid/β−Ti/B2

1050
Temperature (°C)

1000
peritectic liquid/B2/NiTi2

950

eutectic liquid/β−Ti/NiTi2

900
0 5 10 15 20 25 30 35
Fig. 5. Composition of the alloys (solid circles) and of the individual phases Nominal content in Fe (at. )
after homogenization (solid squares) and in the as-cast state (open symbols).
Binary data corresponding to the phases’ composition at 900
C have been Fig. 7. Solidus temperature of the homogenized alloys as a function of iron
plotted with solid diamonds. content.
1230 P. Riani et al. / Intermetallics 14 (2006) 1226e1230
Fig. 8. Microstructure of as-cast samples of alloys A (a) and C (b). Dark grey is b-Ti, intermediate grey is (Fe,Ni)Ti2, and light grey is B2-(Fe,Ni)Ti. The white
precipitates in figure (a) are also associated with this latter phase (see text).
BeF there are clear indications that B2 partly dissolved be-
cause of a peritectic reaction giving (Fe,Ni)Ti2. Other features
observed are that (Fe,Ni)Ti2 grows also dendritically around
the primary B2 phase and that b-Ti deposits as a separate
phase (divorced eutectic) in the last solidifying interdendritic
areas (Fig. 8(b)).
In the case of alloy A, primary dendrites are surrounded
mainly by b-Ti (Fig. 8(a)) which appears to be the second
phase formed during solidification. (Fe,Ni)Ti2 forms rather
irregularly shaped deposits with a smaller size than b-Ti.
Finally, thin and elongated precipitates of a phase with bright
contrast are seen within b-Ti. The composition of this fourth
phase has been reported with an empty circle in Fig. 5. It is
seen that it lies in between the compositions of (Fe,Ni)Ti2 and
B2 phases, so that the value estimated represents certainly
an average of the compositions of these two phases because
of its small size. The shape, size and contrast of this phase sug-
gest that it is the B2-(Fe,Ni)Ti that formed by solid-state pre-
cipitation during cooling.
Fig. 9. Tentative vertical section of the FeeNieTi phase diagram at xTi ¼ 0.66.
The points are the data plotted in Fig. 7.
4. Conclusions
Present results show that the NiTi2 phase is stabilized by
substitution of Fe for Ni, as suggested by the increase in
the solidus temperature with the Fe content (see Fig. 9). Nev-
ertheless, NiTi2 cannot overtake the B2 phase which was
observed to be the primary phase upon solidification in the
whole composition range investigated. This allows to draw
the tentative FeTi2eNiTi2 section of the phase diagram shown
in Fig. 9, according to which the invariant point between liq-
uid, b-Ti, NiTi2 and B2-(Fe,Ni)Ti is much closer to the FeeTi
side than previously suggested [11] and should be of the
U-type: liquid þ (Fe,Ni)Ti 4 b-Ti þ (Fe,Ni)Ti2.
Acknowledgements
This work was performed within the frame of the COST AC-
TION 535 (THALU) and the partial support of COST-ESF and
Italian project MIUR-PRIN2003091235 is acknowledged. The
WDS analyses have been carried out in the ‘‘Service d’analyse
de l’Université P. Sabatier’’, Toulouse, France.
References
[1] Gupta KP. Phase diagrams of ternary nickel alloys, part 1. Indian Institute
of Metals; 1990. p. 321e43.
[2] Vogel R, Wallbaum HS. Arch Eisenhuttenwesen 1938;12:299e304.
[3] Speich GR. Trans Met Soc AIME 1963;227:754e62.
[4] Van Loo FJJ, Vrolifk JVGA, Bastin GF. J Less Common Met 1981;77:
121e30.
[5] Cacciamani G, De Keyzer J, Ferro R, Klotz U, Lacaze J, Wollants P .
Intermetallics 2006 (doi:10.1016/intermet.2005.11.028).
[6] Lee BJ. Calphad 1993;17:251e68.
[7] Hari Kumar KC. XXVIII CALPHAD meeting. Grenoble; 1999.
[8] Bellen P, Hari Kumar KC, Wollants P. Z Metallkd 1996;87:972e8.
[9] Dudkina LP, Kornilov II. Russ Metall 1967;4:98e101.
[10] Abramycheva NL, V’yunitskii IV, Kalmykov KB, Dunaev SF. Vestn
Mosk Univ Ser 2 Khim 1999;40:139e43.
[11] Gupta KP. J Phase Equilib 2001;22:171e5.
[12] Alisova SP, Budberg PB, Barmina TI, Lutskaya NV. Russ Metall
1994;1:117e21.
[13] Efimenko LP, Petrova LP, Sviridov SI. Russ Metall 2000;3:42e6.